You Ask, I Answer: Safe Artificial General Intelligence?

In today's episode, Robert asks how worried should we be about artificial general intelligence? Is there such a thing as safe general intelligence? Is there something such a thing as safe superintelligence? What should we be doing about this? How much should we worry about it?

Don't. Not right now. Worry about what's happening now first. This is something I've seen a lot on LinkedIn too. People worried about super intelligence, a machine that is vastly more intelligent than human beings.

These things are not sentient. They're not self-aware. They're not Skynet. They're not going to create an army of terminators. We don't have the compute power for that.

We do not have the computation power to pull that off. And until we get either bioneural computing or quantum computing working well, and I don't want to work particularly well at the moment, it's not going to happen. When people wax poetic about the dangers of AI, they're often talking about some mythical superintelligent machine deity that's going to wipe out humanity. That is improbable. But it is also kind of a smoke screen.

It is also kind of a misdirection. Because it hides the actual dangers of AI today. Things like bias, right? Making systems that reinforce existing biases. That's bad.

That is something that shouldn't be happening. And it absolutely is happening. One of the biggest dangers of AI that is clear, is present, is obvious, it's happening now is income erosion. And by income erosion, I mean any force that erodes your ability to optimize your income, whatever it is that you do for work. This is nothing new, right?

We've had income erosion forever. It's why things like unions exist. Because companies in a capitalist system inherently always want to maximize profits. That is the function of a company. The function of a company is to generate revenue and maximize profits.

That's that's what companies do, the exception being nonprofits. And so as companies fulfill that mandate, they're going to look for ever cheaper ways to get things done. What is the lowest cost possible way to get this thing done so that we can maximize profits? That isn't necessarily inherently amoral or immoral, that is just what companies do. And AI is the latest step in a long progression of this from slavery and indentured servitude, right?

Centuries ago. That was the easiest way to run a highly profitable company. You enslave a bunch of people and you make them work for nothing and you kill them. Um and you do all sorts of horrible things to them to compel them to do what you want. Um, this that is, you know, slavery is what, thousands of years old.

Um, we've only recently figured out that it's that's kind of bad. Humans. Um we've had unequal labor laws, right? We have there's a period of time up until like the last 200 years where it was like, you know what? It's okay for little kids to work in in in dangerous factories or in dangerous jobs, because they had to, in some cases, like in agriculture, and in other cases, because there was no protection.

And again, companies were like, what's the cheapest way we can get this done? Let's get little kids to do the work instead because we can pay them less. Let's have unevil uneven labor laws where different classes of people get different pay. Uh, if your skin is a certain color, you get paid less. And you hire as many people like that to maximize your profits.

Let's pay women 60% of what we pay men. Get the same work out of them. That is profit maximization. Outsourcing, right? And offshowing.

Let's send the jobs overseas or to some other place that has cheaper labor so we can maximize our profits. You would have things like a banana company, Chiquita, hiring death squads to kill off competition. That's that's like was a couple uh decades ago. That wasn't like in the past, that was like last century, the late end of last century. The court case, I think was in the 90s.

That's real, right? AI is just the latest in this long line of ways that companies try to reduce expenses. And income erosion is a real, a real problem for certain types of work. It's going to impact everything, but certain types of work are going to be impacted more than others because companies want to maximize profits. If it takes you four to six weeks to get a piece of content spec'd, built, developed, delivered, reviewed, and published like a blog post.

And it takes you four to six minutes to do the same thing with chat GPT. And the quality isn't any worse. What's the logical thing you're going to do if you're if your mandate is to maximize profits? You are going to take the four to six minute approach. Back at my old the old agency I used to work at, we had a client that was cranking out 150 blog posts a week.

And they were, they had a huge army of freelancers writing okay content. If you were to go back and look at that content today, ChatGPT could do as good or better a job, create more content and way more cheaply. That is income erosion. So those people who are getting paid uh you know 150 to 300 per blog post, you don't need to pay those people anymore for the same lack of quality. And that's the part that I think is really important for people to understand.

This is something that comes up a lot with generative AI. Generative AI does not have to be the best at what it does, it just has to be better than the average because the average is what you pay for right now. When you hire a freelancer to write blog posts for your company, which a lot of companies do, you're not specifying in the job description that you know, they have to have a Pulitzer. Like you just need to work for 50 cents a word or 25 cents a word, or however we can reduce the cost. And in those situations, AI is a cheaper method by far.

It is faster, and the quality properly prompted isn't worse. It's kind of like that old you know, cliche. You don't have to be the fastest runner to evade a bear, you just have to run faster than the slowest runner in your group. The same thing is true for AI. You don't have to be the best in the world, but you do have to be better than AI, and AI is getting better and better and better day by day.

So that income erosion is a major problem with artificial intelligence, particularly generative AI, because it makes producing the same corporate Dreck way faster and way cheaper. And that will erode the income of people who produce corporate direct slower and more expensively. You would hope that, and you know, I think there are indications that generative AI will actually get create better Dreck, which sounds like a terrible band. Um, but it will do so without those people. And then you have to figure out what else are those people going to do for work.

That is the clear and present danger. And part of remediating that danger, which is why no one in the business world likes to talk about it, is acknowledging that perhaps maximization of profit is not the most important thing. It's important, you got to keep the lights on, you got to pay rent, you got to pay your employees. But perhaps it's not the most important thing the company should be doing. That's why you have um purpose-built companies, companies that uh you know, special types of subcompanies, nonprofits, not for profit entities, that are trying to try alternative methods of doing business where they still have to get made earn a profit, like that's a big killer.

They still have to earn a profit, but they don't necessarily have to maximize profits at all costs, including hiring death squads to kill your competitors, and part of that is going to be about using AI thoughtfully and not handing off work that people enjoy doing. But for the most part, the market is going to rationalize around the fact that AI can do better work faster and much, much, much cheaper. So for those folks who are in those fields, like video creation, image creation, music creation, writing, you've got to figure out how to build your personal brand so that you are irreplaceable and a machine would have a very difficult time replicating you and get good at operating the machines so that you can scale your own unique way of doing things. Because one of the things the fundamental truths, particularly in in B2B stuff, is that this is true generally. People, all other things being equal, people prefer to have relationships with and buy from other people.

That is a window of opportunity if you are a creator in these categories to reduce the impact of income erosion. If people are buying from you because of your brand that they can't get elsewhere, that's an advantage. You've got to take advantage of that. But safe superintelligence, that is to me almost a distraction, an intentional distraction away from the very real challenges that AI is posing today and that we need to address today. If you enjoyed this video, please hit the like button.

Subscribe to my channel if you haven't already. And if you want to know when new videos are available, hit the bell button to be notified as soon as new content is live.